USB disk "disconnect" during very long processing

SaintDanBert · 09-23-2013, 09:12 AM

How do I prevent "timeout" or similar that causes a rotating USB drive to "disconnect" in the middle of a long process? I'm sure that something thinks the workstation is idle because there are no keystrokes or mouse droppings or similar but I'm dashed if I can find the cause of the spin-down.

A typical "long process" might be a workstation backup run, catalog of drive contents, scan drive for duplicate files, and such. These are the sort of jobs one starts at 10pm and leaves to run overnight. I often find a job failure because the drive went "not available" in the middle of the night.

I have a mountain of external drives that connect by USB. Most are "archival" -- documents, photos, etc that I want to keep for a long time. Some are "failsafe" -- backup copies of workstation drives held against catastrophe recovery.

Linux Mint-12 on a Thinkpad notebook.
~~~ 0;-Dan

unSpawn · 09-24-2013, 08:40 PM

Are there any BIOS ACPI or other "sleep" / "wake" settings involved?
Does this happen with each and every external USB casing (wrt chipsets)?
Have you tried USB kernel module debug mode to find more clues? (Watch log size!!!)

SaintDanBert · 09-25-2013, 09:08 AM

Quote:

Originally Posted by unSpawn

Are there any BIOS ACPI or other "sleep" / "wake" settings involved?

I have all of the desktop settings turned off when on AC power.

Quote:

Originally Posted by unSpawn

Does this happen with each and every external USB casing (wrt chipsets)? Have you tried USB kernel module debug mode to find more clues? (Watch log size!!!)

I have not knowledge or experience with this debug mode, but I'm willing to try it. Can you steer me toward HOWTO?

Thanks,
~~~ 0;-Dan

michaelk · 09-25-2013, 09:24 AM

Could be the drive itself going into standby mode. Seagate's drives have / had an auto spindown mode that was problematic with linux. Take a look at the output of the sdparm command.

sdparm -a /dev/sdx (Of course change x for your USB drive)

SaintDanBert · 09-25-2013, 12:33 PM

Quote:

Originally Posted by michaelk

Could be the drive itself going into standby mode. Seagate's drives have / had an auto spindown mode that was problematic with linux. Take a look at the output of the sdparm command.

sdparm -a /dev/sdx (Of course change x for your USB drive)

I tried this:

Code:

prompt$ sudo sdparm --all /dev/sdb
/dev/sdb: Seagate   FreeAgent GoFlex   214

That was not very helpful. I then tried:
(Make sure that you scroll right)

Code:

prompt$ sudo sdparm --all --long --verbose /dev/sdb
>>> about to open device name: /dev/sdb

    /dev/sdb: Seagate   FreeAgent GoFlex   214
    Direct access device specific parameters: WP=0  DPOFUA=0

>> Read write error recovery [rw] mode page [0x1] not found
>> Disconnect-reconnect (SPC + transports) [dr] mode page [0x2] not found
>> Format (SBC) [fo] mode page [0x3] not found
>> Rigid disk (SBC) [rd] mode page [0x4] not found
>> Verify error recovery (SBC) [ve] mode page [0x7] not found
>> Caching (SBC) [ca] mode page [0x8] not found
>> Control [co] mode page [0xa] not found
>> Control extension [coe] mode subpage [0xa,0x1] not found
>> Application tag (SBC) [atag] mode subpage [0xa,0xf0] failed
>> SAT pATA control [pat] mode subpage [0xa,0xf1] not found
>> Power condition - old version [poo] mode page [0xd] not found
>> XOR control (SBC) [xo] mode page [0x10] not found
>> Protocol specific logical unit [pl] mode page [0x18] not found
>> Protocol specific port [pp] mode page [0x19] not found
>> Power condition [po] mode page [0x1a] not found
>> SAT ATA Power condition [apo] mode subpage [0x1a,0xf1] not found
>> Informational exceptions control [ie] mode page [0x1c] not found
>> Background control (SBC) [bc] mode subpage [0x1c,0x1] not found
>> Logical block provisioning (SBC) [lbp] mode subpage [0x1c,0x2] failed

Ahhh!! Much more detail.

I have no idea what this is trying to tell me.
~~~ 8d;-/ Dan

michaelk · 09-25-2013, 01:02 PM

It would be the data contained in the power condition mode page according to some old information from a web search.

SaintDanBert · 09-30-2013, 10:26 AM

Quote:

Originally Posted by michaelk

It would be the data contained in the power condition mode page according to some old information from a web search.

I parse your tokens, but sadly have no idea what they mean...
How might I expose the "power condition mode page"?
How do I make sense out of whatever "data" I find there?
How would I detect that there was a drive setting or condition that causes it to disconnect as I described?

Thanks,
~~~ 0;-Dan